Predicting gallstone pancreatitis in HIV infected patients.

ABSTRACT

BACKGROUND: Human immunodeficiency virus (HIV) infection, low cluster of differentiation (CD)4 counts and antiretroviral therapy can cause cholestasis and raised transaminases. In acute pancreatitis, this may render biochemical predictors of a gallstone aetiology inaccurate. METHODS: In a prospective observational study, acute pancreatitis was diagnosed by standard criteria. Cholecystolithiasis and bile duct diameter were diagnosed by ultrasound. Cholestasis was defined as two of the following bilirubin ≥ 21 umol/l, γ glutamyl transferase ≥ 78 U/l, alkaline phosphatase ≥ 121 U/l. Cholangitis was defined as cholestasis and any two sepsis criteria (temperature > 38˚C, WCC > 12.6 ×109/L, pulse > 90 beats/min). Cholangitis, cholestasis, and bile duct diameter greater that 1 cm were indications for endoscopic retrograde cholangiopancreatography (ERCP). These parameters' ability to predict gallstone pancreatitis (GSP) and choledocholithiasis were compared in HIV+ve and HIV-ve patients. RESULTS: Sixty-two (26%) of 216 patients had GSP. Twenty four were HIV+ve patients. More HIV+ve patients had cholestasis (p = 0.059) and ERCP (p = 0.004). In HIV+ve patients alanine aminotransferase (ALT) > 100 U/L, gamma glutamyl transferase (GGT) > 2 upper limit of normal and cholestasis had a negative predictive value of 92%, 96.7% and 95.2% respectively. In HIV-ve patients, negative predictive value (NPV) was 84%, 83.8% and 84.6% respectively. Bile duct stones were demonstrated at ERCP in 6 (25%) and 3 (8%) of HIV+ve and HIV-ve patients respectively (p = 0.077). Five of 14 ERCP patients had no bile duct stones. HIV+ve and HIV-ve groups had two deaths each. CONCLUSION: Absence at presentation of the abnormal parameters analysed were good predictors of a non-gallstone aetiology particularly in HIV+ve patients. Prior, magnetic resonance cholangiopancreatography (MRCP) or endoscopic ultrasound (EUS) would reduce the number of non-therapeutic ERCPs.